3.1 Linguistic Inquiry and Word Count

Word classification used the Linguistic Inquiry and Word Count 2015 (LIWC) dictionary (Pennebaker et al., 2015) and tokenising tools from the Quanteda R package (Benoit, 2018). Word counts and words-per-sentence were calculated using stringi (Gagolewski, 2018), a wrapper to the ICU regex library.

Following research by Yu (2014) and Newman, Groom, Handelman, & Pennebaker (2008) on gender differences in language, we compared MPs speeches using the following LIWC categories:

• All Pronouns (pronoun)
• First person singular pronouns (i)
• First person plural pronouns (we)
• Verbs (verb)
• Auxiliary verbs (auxverb)
• Social processes (social)
• Positive emotions (posemo)
• Negative emotions (negemo)
• Tentative words (tentat)
• Articles (article)
• Prepositions (preps)
• Anger words (anger)
• Swear words (swear)
• Cognitive processes (cogproc)
• Words longer than six letters (Sixltr)

We also included mean words-per-sentence (WPS), total speach word count (WC) and Flesch–Kincaid grade level (FK) (Kincaid, Fishburne, Rogers, & Chissom, 1975), calculated using Quanteda (Benoit, 2018) and stringi (Gagolewski, 2018).

3.1.2 Shortlists vs Non-Shortlists

The following plots show changes in the occurences of selected LIWC terms, words-per-sentence, total word count and Flesch–Kincaid grade level, over the course of an MP’s career. There do not appear to be any notable changes in speaking style over the course of female Labour MPs’ careers.

Occurence of selected LIWC terms

Effect Sizes for Female Labour MPs by selection process

	All Women Shortlists		Open Shorlists		Effect Size
	Mean	SD	Mean	SD	Cohen’s D	Magnitude
All Pronouns	10.01	4.66	10.19	4.47	-0.04	negligible
First person singular pronouns	1.86	2.41	1.95	2.42	-0.04	negligible
First person plural pronouns	0.88	1.36	1.16	1.51	-0.19	negligible
Verbs	12.88	5.10	12.69	4.80	0.04	negligible
Auxiliary verbs	7.94	3.49	7.86	3.38	0.02	negligible
Social processes	8.46	4.93	8.44	4.58	0.00	negligible
Positive emotions	2.69	2.52	2.81	2.42	-0.05	negligible
Negative emotions	1.17	1.69	1.13	1.68	0.02	negligible
Tentative words	1.48	1.75	1.49	1.73	0.00	negligible
More than six letters	10.56	3.72	10.74	3.56	-0.05	negligible
Articles	7.69	3.38	7.55	3.14	0.04	negligible
Prepositions	12.55	4.54	12.63	4.14	-0.02	negligible
Anger words	0.23	0.79	0.24	0.91	-0.01	negligible
Swear words	0.00	0.06	0.00	0.05	0.01	negligible
Cognitive processes	8.59	4.90	8.85	4.69	-0.05	negligible
Words per Sentence	44.39	20.69	43.21	18.31	0.06	negligible
Total Word Count	401.70	704.15	404.73	664.87	0.00	negligible
Flesh-Kincaid Grade Level	11.13	8.06	10.64	7.15	0.07	negligible

There are no categories among female Labour MPs by selection process meeting the \(|0.2|\) threshold. Only one category – first person plural pronouns, d=0.19 – exceeded \(|0.1|\).

3.2 POS Analysis

Part-of-speech (POS) tagging was done using spaCy (Honnibal & Montani, 2017) and the spacyr package (Benoit & Matsuo, 2018). There is one small gender difference (d = \(|0.22|\)) in the use of plural nouns, which make up 5.86% of the words used by female Labour MPs, compared to 5.04% of words spoken by male Labour MPs. As with LIWC, there are no categories where d >= \(|0.2|\) when comparing female Labour MPs by selection process.

3.3 Tokenising / Keyness

The most commonly used words by both men and women would be protocol decorum expressions, so we calculate the keyness of words to identify gender differences in the choices of topics raised by men and women, and by short-list and non-shortlist women.

3.3.1 Men vs Women

Keyness – a linguistic measure of the frequency of different words in two groups of texts – reveals clear gender differences in the most disproportionately common words used by female and male Labour MPs. Unsurprisingly, despite male MPs saying almost twice as many words (65,827,133 vs 34,159,304) as their female colleagues, female Labour MPs were more than two-and-a-half (2.61) times as likely to say “women”. They were also much more likely to use “women’s” and “woman” in parliamentary debate. Female Labour MPs also appear much more likely to discuss “children”, “people”, “care”, “families”, “home”, “parents”, “work” and social policy areas such as “services”, “disabled [people]” and “housing” than their male colleagues. Male MPs were more likely to refer to military topics (“Iraq”, “nuclear”), and to parliamentary process and protocol – “question”, “political”, “conservative”, “electoral”, “house”, “party”, “argument” “liberal” and “point” are far more common in speeches by male Labour MPs than by female ones. This could suggest that male Labour MPs are more comfortable using the traditional language of House of Commons debate, and are more concerned with the rules, procedures and processes of the parliamentary system than their female colleagues.

Keyness between Labour MPs, by Gender

3.3.2 Shortlists vs Non-Shortlists

Keyness differences by selection process are not as obviously stereotypical. Nonetheless, the most common words amongst AWS MPs included “carers”, “disabled”, “bedroom” and “sen” (Special Educational Needs). Also of note is AWS MPs making more references to their “constituency” and its “constituents”, suggesting that AWS MPs may draw on the fact they were elected by their constituents as a source political legitimacy, at least more than non-AWS MPs.

Keyness between Female Labour MPs, by Selection Process

3.3.3 Labour vs Conservative

The keyness differences between Labour and Conservative MPs are much greater than gender differences within Labour. The very high use of “Lady” by Conservative MPs is reflective of the greater proportion of female MPs in other parties, as it is often used to refer to comments by other members of the house. It may also represent a greater use of traditional hosue decorum by Conservative MPs.

Keyness between Labour and Conservative MPs

3.4 Bigrams

We created bigrams of all first person plural and singular pronouns for female Labour MPs. As above, AWS MPs are far more likely to make references to their constituency or their constituents.

Bigram Keyness in Female Labour MPs by Selection Process

3.5 Naive Bayes classification

We trained a Naive Bayes classifier with document-frequency priors and a multinomial distribution to predict the gender of speakers when given speeches by all Labour MPs in our dataset, and the selection process when only given female Labour MPs. The accuracy of both models were roughly equivalent, 70.67% accuracy when predicting gender and 71.22% when predicting shortlists. By contrast, the classifier could distinguish between Labour and Conservative speeches with 74.23% accuracy.

3.6 Topic Models

Using topic models to classify text is widely used in social sciences (Grimmer & Stewart, 2013), as, when combined with the large volume of plain text data available, it allows for a rapid and consistent method of analysis . Topic modelling and other statistic methods of textual analysis are not a substitute for reading the texts themselves, but can augment other analysis or – as in this case – analyse and classify larger amounts of text than would be feasible using human coders (Grimmer & Stewart, 2013). Topic models classify a series of documents (in this case individual speeches) into one of a given number of topics, identifying terms that are common in some documents but rare in others. When developing topic models, there is a trade-off between high precision in the classification of each document with broader topics when using smaller numbers of topics, or lower precision in individual speech classification with more finely-grained topics when using larger numbers of topics. Grimmer & Stewart (2013) also highlight the importance of validating unsurpervised topic models when applied to new sets of texts.

The R package stm (Roberts, Stewart, & Tingley, 2018) implements a structured topic model (STM) (Arora et al., 2013; Roberts, Stewart, & Airoldi, 2016). An STM incorporates data about the writer or speaker into the topic classification algorithm. This differs from traditional topic modelling methods using latent variables to identify topics (e.g. with latent Dirichlet allocation Blei, Ng, & Jordan, 2003), and then comparing proportions of each topic to one or more external variables. STM allows us to incorporate the variables we are interested in to the topic model itself using a generalised linear model; i.e. the proportion of speechs classified as belonging to each topic can vary as a function of the AWS and gender variables.

We incorporated the AWS status of speakers and their gender as covariates into our topic model, using all speeches by Labour MPs.

Topic Model Selection

To identify the best number of topics, we created six topic models with different numbers of topics (K). We created models with 30, 45, 60, 80 and 100 topics, and used an algorithm developed by Lee & Mimno (2014), implemented in the stm package (Roberts et al., 2018), which resulted in K = 66. As seen in Figure , the algorithm from Lee & Mimno (2014) appears to produce the best result, a topic model with 66 topics, across 251,072 speeches with a dictionary of 241,625 words. All models were created using the “spectral” method developed by Arora et al. (2013), implemented by Roberts et al. (2018). One of the topics – Topic 66 – was identified as a possibly topic, but was never the most likely topic in the matrix of number of documents by number of topics – labelled \(\theta\) by Roberts et al. (2018), and so while it is included in the model, assignment of single topics to speeches uses the highest \(\theta\) for each speech. Other topics are rarely used – Topic 53, which we labelled “Dispatch Box”, only has five topics assigned to it, four from Male MPs and one from an AWS MP.

3.6.1 Shortlists vs Non-Shortlists

We created a Fruchterman-Reingold (Fruchterman & Reingold, 1991) diagram (Figure ) to show the connections between different topics. Larger vertices indicate more common topics, and the plot uses a colour scale to indicate the proportion of speeches classed in that topic made by AWS and non-AWS female Labour MPs. The space between vertices indicate the closeness or distance of two topics. E.g. closer vertices represent topics with more overlapping words than more distant topics.

Fruchterman-Reingold plot of k0 Network

Topic Estimates

	Estimate	Standard Error	t value	Pr(>|t|)
Topic 1 – Employment & unions
Intercept	0.0117293	0.0003265	35.9229375	< 0.001	***
Shortlist	-0.0010013	0.0003917	-2.5560852	0.011
Men	0.0003337	0.0003501	0.9533037	0.34
Topic 2 – Legal system
Intercept	0.0237673	0.0004426	53.7032696	< 0.001	***
Shortlist	-0.0103830	0.0004923	-21.0892490	< 0.001	***
Men	-0.0070596	0.0004687	-15.0610097	< 0.001	***
Topic 3 – Roads
Intercept	0.0102028	0.0003492	29.2159550	< 0.001	***
Shortlist	-0.0004374	0.0004409	-0.9920033	0.32
Men	0.0014442	0.0003970	3.6376843	< 0.001	***
Topic 4 – Housing
Intercept	0.0157555	0.0004302	36.6205329	< 0.001	***
Shortlist	0.0015909	0.0005400	2.9460710	0.003	**
Men	-0.0044874	0.0004786	-9.3768098	< 0.001	***
Topic 5 – Police, firefighters & prison
Intercept	0.0173574	0.0004351	39.8969615	< 0.001	***
Shortlist	-0.0036291	0.0005530	-6.5624923	< 0.001	***
Men	-0.0032971	0.0004727	-6.9753678	< 0.001	***
Topic 6 – Northern Ireland
Intercept	0.0090221	0.0001120	80.5664090	< 0.001	***
Shortlist	-0.0004340	0.0001343	-3.2306114	0.001	**
Men	-0.0000588	0.0001205	-0.4875685	0.63
Topic 7 – Committee
Intercept	0.0205896	0.0003505	58.7503260	< 0.001	***
Shortlist	-0.0012582	0.0004192	-3.0014762	0.003	**
Men	0.0007352	0.0003663	2.0074731	0.045
Topic 8 – Schools
Intercept	0.0137290	0.0004676	29.3601167	< 0.001	***
Shortlist	0.0030492	0.0005829	5.2306431	< 0.001	***
Men	0.0010367	0.0005043	2.0557566	0.040
Topic 9 – Energy & climate change
Intercept	0.0158405	0.0005106	31.0203384	< 0.001	***
Shortlist	-0.0023065	0.0006084	-3.7913302	< 0.001	***
Men	0.0011314	0.0005556	2.0362465	0.042
Topic 10 – Defence
Intercept	0.0081750	0.0004488	18.2132671	< 0.001	***
Shortlist	0.0022589	0.0005207	4.3380074	< 0.001	***
Men	0.0076142	0.0004699	16.2027698	< 0.001	***
Topic 11 – Parliament
Intercept	0.0081994	0.0001755	46.7284520	< 0.001	***
Shortlist	0.0026092	0.0002237	11.6662958	< 0.001	***
Men	0.0036992	0.0001904	19.4256979	< 0.001	***
Topic 12 – International politics
Intercept	0.0084607	0.0003026	27.9560687	< 0.001	***
Shortlist	-0.0013337	0.0003621	-3.6835860	< 0.001	***
Men	0.0041934	0.0003371	12.4396572	< 0.001	***
Topic 13 – Ministers
Intercept	0.0138491	0.0002584	53.5943814	< 0.001	***
Shortlist	0.0060202	0.0003288	18.3076055	< 0.001	***
Men	0.0028971	0.0002826	10.2500492	< 0.001	***
Topic 14 – Policy impact
Intercept	0.0118057	0.0001146	103.0353298	< 0.001	***
Shortlist	0.0010850	0.0001425	7.6121479	< 0.001	***
Men	-0.0002660	0.0001210	-2.1989509	0.028
Topic 15 – Gender
Intercept	0.0173153	0.0003925	44.1143113	< 0.001	***
Shortlist	-0.0003599	0.0004682	-0.7686723	0.44
Men	-0.0124314	0.0003951	-31.4628115	< 0.001	***
Topic 16 – Regional development
Intercept	0.0302478	0.0002819	107.3185086	< 0.001	***
Shortlist	-0.0069433	0.0003475	-19.9787322	< 0.001	***
Men	-0.0072041	0.0003139	-22.9512530	< 0.001	***
Topic 17 – Communications
Intercept	0.0090915	0.0002875	31.6210394	< 0.001	***
Shortlist	-0.0005277	0.0003290	-1.6038348	0.11
Men	0.0006568	0.0003074	2.1366737	0.033
Topic 18 – Immigration
Intercept	0.0094290	0.0002129	44.2879525	< 0.001	***
Shortlist	-0.0011745	0.0002571	-4.5690616	< 0.001	***
Men	-0.0007350	0.0002316	-3.1735568	0.002	**
Topic 19 – Health system
Intercept	0.0275233	0.0005157	53.3703952	< 0.001	***
Shortlist	-0.0049952	0.0006407	-7.7962889	< 0.001	***
Men	-0.0113568	0.0005738	-19.7924994	< 0.001	***
Topic 20 – International development
Intercept	0.0165058	0.0004885	33.7876752	< 0.001	***
Shortlist	-0.0038420	0.0005626	-6.8292777	< 0.001	***
Men	-0.0004560	0.0005183	-0.8798630	0.38
Topic 21 – Benefits & disability
Intercept	0.0129030	0.0003794	34.0065782	< 0.001	***
Shortlist	0.0111579	0.0004627	24.1141591	< 0.001	***
Men	-0.0008280	0.0004026	-2.0565363	0.040
Topic 22 – Sport & culture
Intercept	0.0102223	0.0003471	29.4528191	< 0.001	***
Shortlist	0.0031663	0.0004470	7.0833239	< 0.001	***
Men	0.0024897	0.0003954	6.2974174	< 0.001	***
Topic 23 – History
Intercept	0.0076529	0.0002277	33.6140265	< 0.001	***
Shortlist	0.0021342	0.0002825	7.5542636	< 0.001	***
Men	0.0060746	0.0002510	24.2024434	< 0.001	***
Topic 24 – Higher education & skills
Intercept	0.0133129	0.0004092	32.5311711	< 0.001	***
Shortlist	0.0008458	0.0005069	1.6685611	0.095
Men	0.0010407	0.0004441	2.3434975	0.019
Topic 25 – Concurring point
Intercept	0.0135965	0.0001095	124.1478785	< 0.001	***
Shortlist	-0.0011169	0.0001369	-8.1566463	< 0.001	***
Men	0.0018820	0.0001199	15.6905274	< 0.001	***
Topic 26 – Pensions
Intercept	0.0153734	0.0003769	40.7921511	< 0.001	***
Shortlist	0.0020502	0.0004754	4.3123169	< 0.001	***
Men	-0.0006438	0.0004162	-1.5466952	0.12
Topic 27 – Points of order
Intercept	0.0112390	0.0002969	37.8547674	< 0.001	***
Shortlist	0.0017422	0.0003435	5.0713744	< 0.001	***
Men	0.0065973	0.0003179	20.7512698	< 0.001	***
Topic 28 – Issues
Intercept	0.0415230	0.0002561	162.1644690	< 0.001	***
Shortlist	-0.0096543	0.0003259	-29.6238529	< 0.001	***
Men	-0.0070418	0.0002740	-25.7041381	< 0.001	***
Topic 29 – Constituencies
Intercept	0.0142783	0.0001244	114.7998010	< 0.001	***
Shortlist	0.0018606	0.0001599	11.6386483	< 0.001	***
Men	-0.0010999	0.0001399	-7.8613191	< 0.001	***
Topic 30 – Ethnic groups & racism
Intercept	0.0105215	0.0002009	52.3711531	< 0.001	***
Shortlist	-0.0000386	0.0002463	-0.1565624	0.88
Men	-0.0019437	0.0002127	-9.1370020	< 0.001	***
Topic 31 – Amendments
Intercept	0.0132683	0.0003943	33.6518657	< 0.001	***
Shortlist	-0.0016187	0.0005022	-3.2233676	0.001	**
Men	0.0017199	0.0004485	3.8343595	< 0.001	***
Topic 32 – Reports
Intercept	0.0181610	0.0002731	66.4919467	< 0.001	***
Shortlist	0.0001533	0.0003295	0.4652214	0.64
Men	-0.0011864	0.0002988	-3.9704983	< 0.001	***
Topic 33 – People
Intercept	0.0354711	0.0002747	129.1097920	< 0.001	***
Shortlist	0.0012343	0.0003350	3.6840510	< 0.001	***
Men	0.0022768	0.0003020	7.5394479	< 0.001	***
Topic 34 – Wales & Scotland
Intercept	0.0087640	0.0003532	24.8115542	< 0.001	***
Shortlist	0.0024673	0.0004284	5.7594006	< 0.001	***
Men	0.0047894	0.0003815	12.5533013	< 0.001	***
Topic 35 – Alcohol & tobacco
Intercept	0.0099309	0.0003934	25.2461470	< 0.001	***
Shortlist	0.0022555	0.0004723	4.7755180	< 0.001	***
Men	0.0009243	0.0004249	2.1752110	0.030
Topic 36 – Place names
Intercept	0.0084177	0.0001683	50.0262769	< 0.001	***
Shortlist	0.0011394	0.0001943	5.8631456	< 0.001	***
Men	-0.0000611	0.0001809	-0.3378675	0.74
Topic 37 – Budget
Intercept	0.0223751	0.0004285	52.2169657	< 0.001	***
Shortlist	0.0029730	0.0005221	5.6940210	< 0.001	***
Men	0.0022528	0.0004679	4.8144816	< 0.001	***
Topic 38 – Tax
Intercept	0.0179353	0.0004635	38.6935102	< 0.001	***
Shortlist	0.0069110	0.0005804	11.9075354	< 0.001	***
Men	0.0014154	0.0005128	2.7602415	0.006	**
Topic 39 – Private companies
Intercept	0.0129346	0.0003251	39.7926296	< 0.001	***
Shortlist	-0.0024247	0.0003749	-6.4679740	< 0.001	***
Men	-0.0005574	0.0003443	-1.6191062	0.11
Topic 40 – Environment & fishing
Intercept	0.0063004	0.0003159	19.9435303	< 0.001	***
Shortlist	0.0010055	0.0003948	2.5468657	0.011
Men	0.0031572	0.0003469	9.1009323	< 0.001	***
Topic 41 – Crime
Intercept	0.0227176	0.0004926	46.1214787	< 0.001	***
Shortlist	-0.0052100	0.0006485	-8.0334463	< 0.001	***
Men	-0.0086317	0.0005406	-15.9673915	< 0.001	***
Topic 42 – Bills
Intercept	0.0265455	0.0004373	60.7092286	< 0.001	***
Shortlist	-0.0050558	0.0005059	-9.9927473	< 0.001	***
Men	-0.0021254	0.0004679	-4.5421291	< 0.001	***
Topic 43 – Children
Intercept	0.0168410	0.0003635	46.3335674	< 0.001	***
Shortlist	0.0004441	0.0004285	1.0363559	0.30
Men	-0.0091613	0.0003643	-25.1467096	< 0.001	***
Topic 44 – Utilities & PFI
Intercept	0.0115949	0.0002094	55.3799410	< 0.001	***
Shortlist	0.0010239	0.0002797	3.6601509	< 0.001	***
Men	0.0007618	0.0002291	3.3254506	< 0.001	***
Topic 45 – Middle East
Intercept	0.0146987	0.0005523	26.6126307	< 0.001	***
Shortlist	0.0009502	0.0006461	1.4706954	0.14
Men	0.0027723	0.0005999	4.6214373	< 0.001	***
Topic 46 – Local authorities
Intercept	0.0222754	0.0003889	57.2794054	< 0.001	***
Shortlist	-0.0041265	0.0004790	-8.6149378	< 0.001	***
Men	-0.0042817	0.0004210	-10.1706503	< 0.001	***
Topic 47 – Elections
Intercept	0.0089868	0.0003732	24.0834565	< 0.001	***
Shortlist	0.0022867	0.0004628	4.9407579	< 0.001	***
Men	0.0092117	0.0004143	22.2366792	< 0.001	***
Topic 48 – Debate
Intercept	0.0145013	0.0001694	85.5989488	< 0.001	***
Shortlist	0.0024682	0.0002201	11.2138470	< 0.001	***
Men	0.0035042	0.0001820	19.2564073	< 0.001	***
Topic 49 – Transport
Intercept	0.0136513	0.0004674	29.2047554	< 0.001	***
Shortlist	0.0037373	0.0005495	6.8015254	< 0.001	***
Men	0.0028003	0.0005334	5.2503520	< 0.001	***
Topic 50 – Questions
Intercept	0.0162856	0.0001742	93.4755707	< 0.001	***
Shortlist	0.0001110	0.0002273	0.4881722	0.63
Men	-0.0001163	0.0001918	-0.6065259	0.54
Topic 51 – Families
Intercept	0.0120526	0.0003133	38.4753026	< 0.001	***
Shortlist	0.0039033	0.0003881	10.0567601	< 0.001	***
Men	-0.0019514	0.0003235	-6.0318480	< 0.001	***
Topic 52 – Health research
Intercept	0.0165420	0.0004632	35.7104048	< 0.001	***
Shortlist	-0.0041452	0.0005768	-7.1861974	< 0.001	***
Men	-0.0077711	0.0004907	-15.8373261	< 0.001	***
Topic 53 – Dispatch box
Intercept	0.0064128	0.0000583	110.0515217	< 0.001	***
Shortlist	0.0001883	0.0000754	2.4969121	0.013
Men	0.0011404	0.0000649	17.5829883	< 0.001	***
Topic 54 – Parties
Intercept	0.0182173	0.0003294	55.3061719	< 0.001	***
Shortlist	0.0006025	0.0003943	1.5282419	0.13
Men	0.0065954	0.0003576	18.4412214	< 0.001	***
Topic 55 – Statements
Intercept	0.0166560	0.0001709	97.4505160	< 0.001	***
Shortlist	0.0029678	0.0002089	14.2036543	< 0.001	***
Men	0.0045004	0.0001845	24.3869846	< 0.001	***
Topic 56 – European Union
Intercept	0.0138231	0.0004121	33.5403197	< 0.001	***
Shortlist	-0.0028693	0.0004899	-5.8575016	< 0.001	***
Men	0.0025373	0.0004589	5.5288586	< 0.001	***
Topic 57 – Locations
Intercept	0.0076035	0.0002284	33.2860565	< 0.001	***
Shortlist	0.0024954	0.0002981	8.3698996	< 0.001	***
Men	0.0024334	0.0002469	9.8548887	< 0.001	***
Topic 58 – Jobs & manufacturing
Intercept	0.0158321	0.0004535	34.9095464	< 0.001	***
Shortlist	0.0029878	0.0005815	5.1376590	< 0.001	***
Men	0.0017569	0.0004690	3.7461544	< 0.001	***
Topic 59 – Small business
Intercept	0.0076018	0.0001803	42.1695772	< 0.001	***
Shortlist	-0.0008847	0.0002135	-4.1429385	< 0.001	***
Men	-0.0005124	0.0001926	-2.6603523	0.008	**
Topic 60 – Agreement & disagreement
Intercept	0.0239422	0.0002724	87.8797300	< 0.001	***
Shortlist	-0.0019971	0.0003095	-6.4532664	< 0.001	***
Men	0.0089018	0.0002907	30.6202123	< 0.001	***
Topic 61 – Voluntary sector
Intercept	0.0297649	0.0003695	80.5462637	< 0.001	***
Shortlist	-0.0053496	0.0004420	-12.1031545	< 0.001	***
Men	-0.0110179	0.0003997	-27.5681990	< 0.001	***
Topic 62 – Comments
Intercept	0.0123738	0.0001540	80.3723165	< 0.001	***
Shortlist	-0.0011508	0.0001876	-6.1327653	< 0.001	***
Men	0.0028889	0.0001684	17.1576417	< 0.001	***
Topic 63 – Social care
Intercept	0.0185639	0.0003583	51.8090125	< 0.001	***
Shortlist	-0.0021038	0.0004626	-4.5476135	< 0.001	***
Men	-0.0094873	0.0003931	-24.1315296	< 0.001	***
Topic 64 – Time
Intercept	0.0192974	0.0001794	107.5620807	< 0.001	***
Shortlist	0.0004407	0.0001988	2.2172156	0.027
Men	0.0020854	0.0001932	10.7939903	< 0.001	***
Topic 65 – Media & animals
Intercept	0.0063578	0.0003538	17.9711224	< 0.001	***
Shortlist	0.0039731	0.0004477	8.8753280	< 0.001	***
Men	0.0057876	0.0003792	15.2628369	< 0.001	***
Topic 66 – Other
Intercept	0.0040907	0.0000274	149.5004561	< 0.001	***
Shortlist	0.0000454	0.0000369	1.2276619	0.22
Men	-0.0002613	0.0000295	-8.8644823	< 0.001	***

Count and Distribution of Topics – k0

Topic	AWS Speeches	Percent of AWS Speeches	Non-AWS Speeches	Percent of non-AWS Speeches	Male MP Speeches	Percent of Male MP Speeches
Employment & unions	453	0.84%	260	0.93%	2,154	1.27%
Legal system	866	1.61%	1,098	3.94%	3,882	2.29%
Roads	558	1.04%	298	1.07%	2,141	1.26%
Housing	1,384	2.57%	664	2.38%	2,416	1.43%
Police, firefighters & prison	1,046	1.94%	710	2.55%	3,349	1.98%
Northern Ireland	221	0.41%	66	0.24%	604	0.36%
Committee	1,049	1.95%	492	1.77%	3,885	2.29%
Schools	1,368	2.54%	522	1.87%	3,780	2.23%
Energy & climate change	1,105	2.05%	746	2.68%	4,628	2.73%
Defence	793	1.47%	280	1%	3,999	2.36%
Parliament	376	0.7%	85	0.31%	1,081	0.64%
International politics	290	0.54%	161	0.58%	2,026	1.2%
Ministers	874	1.62%	241	0.86%	2,085	1.23%
Policy impact	242	0.45%	68	0.24%	417	0.25%
Gender	1,258	2.34%	702	2.52%	553	0.33%
Regional development	932	1.73%	709	2.54%	2,712	1.6%
Communications	385	0.72%	287	1.03%	1,742	1.03%
Immigration	422	0.78%	219	0.79%	1,212	0.72%
Health system	2,146	3.99%	1,489	5.34%	4,684	2.76%
International development	861	1.6%	687	2.47%	3,722	2.2%
Benefits & disability	1,890	3.51%	317	1.14%	2,100	1.24%
Sport & culture	846	1.57%	317	1.14%	2,627	1.55%
History	299	0.56%	140	0.5%	1,720	1.02%
Higher education & skills	976	1.81%	455	1.63%	3,503	2.07%
Concurring point	33	0.06%	9	0.03%	136	0.08%
Pensions	1,230	2.29%	527	1.89%	2,980	1.76%
Points of order	788	1.47%	230	0.83%	4,069	2.4%
Issues	1,619	3.01%	1,721	6.18%	6,743	3.98%
Constituencies	125	0.23%	30	0.11%	227	0.13%
Ethnic groups & racism	454	0.84%	203	0.73%	946	0.56%
Amendments	526	0.98%	317	1.14%	2,294	1.35%
Reports	537	1%	322	1.16%	1,491	0.88%
People	2,816	5.24%	1,049	3.76%	9,128	5.39%
Wales & Scotland	661	1.23%	225	0.81%	2,659	1.57%
Alcohol & tobacco	845	1.57%	336	1.21%	2,359	1.39%
Place names	163	0.3%	47	0.17%	446	0.26%
Budget	1,615	3%	665	2.39%	5,568	3.29%
Tax	2,148	3.99%	691	2.48%	4,557	2.69%
Private companies	453	0.84%	363	1.3%	1,795	1.06%
Environment & fishing	434	0.81%	186	0.67%	1,690	1%
Crime	1,409	2.62%	925	3.32%	3,075	1.82%
Bills	1,198	2.23%	931	3.34%	4,538	2.68%
Children	1,174	2.18%	631	2.26%	1,299	0.77%
Utilities & PFI	434	0.81%	175	0.63%	1,418	0.84%
Middle East	1,285	2.39%	587	2.11%	4,539	2.68%
Local authorities	1,051	1.95%	711	2.55%	3,687	2.18%
Elections	758	1.41%	240	0.86%	4,309	2.54%
Debate	423	0.79%	128	0.46%	1,365	0.81%
Transport	1,518	2.82%	547	1.96%	4,169	2.46%
Questions	390	0.73%	182	0.65%	1,116	0.66%
Families	786	1.46%	276	0.99%	1,170	0.69%
Health research	745	1.39%	592	2.12%	1,468	0.87%
Dispatch box	1	0%	NA	NA%	4	0%
Parties	882	1.64%	439	1.58%	5,053	2.98%
Statements	180	0.33%	79	0.28%	860	0.51%
European Union	768	1.43%	554	1.99%	3,945	2.33%
Locations	299	0.56%	126	0.45%	1,111	0.66%
Jobs & manufacturing	1,424	2.65%	587	2.11%	4,158	2.45%
Small business	230	0.43%	183	0.66%	793	0.47%
Agreement & disagreement	524	0.97%	275	0.99%	4,965	2.93%
Voluntary sector	1,303	2.42%	854	3.06%	2,478	1.46%
Comments	108	0.2%	95	0.34%	866	0.51%
Social care	862	1.6%	520	1.87%	1,185	0.7%
Time	208	0.39%	103	0.37%	928	0.55%
Media & animals	741	1.38%	189	0.68%	2,812	1.66%

K0 Pyramid Chart

k0 Bar Chart

3.6.1.1 Word Occurences

The table below shows the twenty most common words in each topic, and the twenty words with the highest FREX score, a measure that uses a harmonic mean of word exclusivity and topic coherence (Airoldi & Bischof, 2016). We have named each topic based on the most common words and highest FREX score words in each topic.

Words in topic - k0

Topic Number	Top Twenty Words	Top Twenty FREX
Employment & unions	rights, workers, law, human, civil, trade, union, employers, protection, act, employment, unions, safety, employees, work, service, staff, employer, legislation, protect	tupe, blacklisting, acas, rights, gangmasters, civil, dispute, protections, unions, dismissal, servants, disputes, human, workers, employer, num, certification, employees, tuc, employers
Legal system	cases, court, legal, case, justice, law, courts, evidence, lord, appeal, system, criminal, judicial, investigation, judge, aid, prosecution, circumstances, trial, lawyers	judicial, attorney-general, court, judges, prosecutor, carlile, defendant, extradition, cps, judiciary, admissible, pre-charge, jury, solicitors, lawyers, solicitor, courts, lawyer, detention, judge
Roads	road, planning, site, land, sites, car, vehicles, residents, roads, safety, use, driving, vehicle, park, development, traffic, drivers, area, cars, speed	bikes, cyclists, pedestrians, gypsy, off-road, cycling, encampments, parking, highways, masts, drivers, belt, roads, highway, road, gypsies, vehicles, site, vehicle, bike
Housing	housing, homes, social, affordable, property, home, properties, london, accommodation, building, private, houses, tenants, rent, need, council, landlords, sector, buy, people	tenants, rent, landlords, rented, homelessness, rents, leaseholders, leasehold, tenancy, commonhold, hmos, housing, one-bedroom, homeless, properties, right-to-buy, affordable, sleepers, fulham, landlord
Police, firefighters & prison	police, officers, crime, policing, service, fire, prison, home, force, chief, community, officer, staff, forces, neighbourhood, probation, prisons, safety, prisoners, resources	policing, firefighters, constables, pcsos, probation, csos, prisons, fire, constable, hmic, constabulary, officers, police, prison, prisoners, reoffending, neighbourhood, metropolitan, fires, ipcc
Northern Ireland	make, sure, progress, northern, decisions, ireland, difference, towards, future, process, contribution, statement, responsibilities, easier, responsibility, must, departmental, belfast, friday, choices	sinn, fein, make, sure, belfast, northern, progress, ulster, difference, ireland, ruc, decisions, patten, dissident, departmental, antrim, taoiseach, imc, chastelain, dpps
Committee	committee, report, review, commission, independent, government, select, process, evidence, inquiry, scrutiny, recommendations, role, board, set, work, reports, public, published, parliament	committee’s, select, inquiry, scrutiny, committee, recommendations, committees, independent, recommendation, panel, pre-legislative, report, chairman, review, reviews, scrutinise, inquiries, conclusions, publication, findings
Schools	schools, school, education, teachers, pupils, primary, children, standards, educational, special, secondary, parents, free, teacher, teaching, head, academies, academy, curriculum, good	schools, teachers, pupils, academies, pupil, grammar, classroom, leas, school’s, academisation, school, teacher, bsf, academy, headteachers, ofsted, literacy, curriculum, lea, classrooms
Energy & climate change	energy, climate, change, fuel, carbon, gas, power, emissions, waste, nuclear, prices, wind, green, environmental, electricity, oil, industry, efficiency, renewable, price	energy, carbon, electricity, renewable, renewables, solar, ofgem, greenhouse, co2, ccs, feed-in, biofuels, microgeneration, fossil, sellafield, decarbonisation, chp, shale, mw, bnfl
Defence	defence, forces, armed, afghanistan, service, military, personnel, army, security, troops, support, ministry, royal, veterans, british, force, capability, iraq, equipment, also	armed, veterans, mod, regiment, legion, servicemen, reservists, helmand, battalion, ta, hms, gurkhas, regiments, marines, gurkha, fusiliers, ex-service, eurofighter, isaf, afghan
Parliament	house, leader, motion, commons, therefore, parliament, petition, parliamentary, government, urge, present, signed, table, notes, library, behalf, remain, floor, westminster, request	petitioners, declares, petition, house, motion, urges, commons, serjeant, recess, notes, leader, motions, lobbyist, thursday, e-petitions, early-day, house’s, tuesday, session, lobbying
International politics	united, states, agreement, kingdom, foreign, treaty, council, security, us, nuclear, president, co-operation, convention, nations, national, policy, article, russia, international, position	lisbon, ratification, treaty, non-proliferation, treaties, qmv, veto, ratified, gibraltar, ukraine, russia, protocol, agreement, states, united, ratify, kingdom’s, russian, hague, disarmament
Ministers	secretary, state, statement, ministers, today, confirm, department, government’s, explain, yesterday, home, plans, announcement, government, welcome, chief, state’s, urgent, ministerial, announced	secretary, state, state’s, confirm, ministers, yesterday, announcement, ministerial, explain, statement, expects, urgent, intends, assurances, yesterday’s, secretaries, secretary’s, update, leaked, cabinet
Policy impact	made, clear, number, decision, impact, changes, recent, assessment, effect, level, discussions, likely, proposed, colleagues, potential, representations, implications, analysis, effects, result	made, clear, decision, assessment, recent, changes, impact, representations, implications, effect, discussions, analysis, assess, implementation, estimate, level, number, negative, outcome, colleagues
Gender	women, men, violence, equality, domestic, age, discrimination, women’s, equal, pay, woman, girls, gender, sexual, sex, female, gap, government, maternity, male	women’s, gender, transgender, breastfeeding, refuges, women, abortions, fgm, shortlists, female, male, equality, girls, all-women, gay, equalities, lesbian, men, pregnancy, fawcett
Regional development	new, development, future, programme, national, strategy, government, regional, key, plan, department, welcome, paper, set, ensure, commitment, support, improve, need, deliver	strategy, regional, programme, projects, paper, plan, project, deliver, white, key, development, delivering, develop, priorities, partnership, improve, framework, new, priority, improving
Communications	office, post, bank, banks, rural, offices, services, service, royal, banking, network, mail, closure, access, areas, broadband, card, staff, account, closures	offices, mail, sub-postmasters, sub-post, superfast, post, postwatch, postcomm, consignia, broadband, rbs, office, mail’s, banking, bank, lloyds, ons, branches, uso, banks
Immigration	british, uk, rules, home, immigration, citizens, asylum, identity, status, country, overseas, application, indicated, applications, apply, border, abroad, cards, migration, entry	passports, nationality, dissent, immigration, passport, indicated, points-based, identity, asylum, nationals, visa, dependencies, migration, migrants, biometric, overseas, citizen, entry, abroad, monarch
Health system	health, nhs, hospital, service, patients, services, mental, trust, staff, hospitals, care, trusts, patient, primary, waiting, doctors, nurses, e, gp, emergency	in-patient, helier, nurses, chcs, nhs, ccgs, ccg, sha, hospital’s, hospital, fundholding, pct, hospitals, mental, gp, healthwatch, orthopaedic, walk-in, trusts, reconfiguration
International development	international, countries, world, aid, development, government, developing, africa, global, uk, support, trade, poverty, country, india, assistance, un, need, also, nations	zimbabwe, dfid, burma, congo, cdc, kenya, burmese, doha, uganda, mugabe, sub-saharan, g8, zimbabwean, dfid’s, gleneagles, african, sri, lanka, cancun, nigeria
Benefits & disability	people, benefit, work, benefits, disabled, support, allowance, welfare, employment, disability, system, government, help, universal, credit, reform, get, vulnerable, plus, living	incapacity, dla, esa, jobcentre, disabled, jobseeker’s, jsa, disability, allowance, dwp, claimants, atos, benefit, plus, claiming, pip, motability, benefits, deaf, bedroom
Sport & culture	city, centre, town, sport, football, community, liverpool, sports, club, constituency, clubs, culture, london, great, facilities, one, bid, games, towns, regeneration	football, olympic, museum, museums, stadium, athletes, cricket, paralympic, games, gospels, sports, club, sporting, fans, cup, rugby, arts, olympics, sport, galleries
History	history, former, world, tribute, great, day, never, proud, first, remember, new, john, campaign, century, parliament, pay, also, war, today, sir	maiden, miners, memorial, predecessors, tony, hillsborough, martin, james, john, william, andrew, margaret, anniversary, peter, alan, fought, memories, harold, churchill, edward
Higher education & skills	education, skills, students, university, training, higher, young, universities, college, learning, science, apprenticeships, colleges, fees, student, funding, research, system, qualifications, courses	universities, student, apprenticeship, fe, graduates, ema, graduate, students, colleges, diploma, apprenticeships, vocational, leitch, esol, qualifications, courses, university, undergraduate, tuition, sixth-form
Concurring point	point, agree, country, making, makes, absolutely, whole, much, good, part, friend’s, entirely, completely, kind, sense, giving, rather, share, precisely, parts	agree, absolutely, makes, friend’s, point, precisely, making, entirely, completely, kind, whole, sense, direction, mentions, refers, gentleman’s, describes, powerful, danger, exactly
Pensions	scheme, pension, credit, pensions, insurance, schemes, pensioners, payments, compensation, fund, payment, money, financial, paid, savings, debt, retirement, government, pay, income	pension, annuity, policyholders, annuities, auto-enrolment, insurance, retirement, loan, payments, payday, scheme, compensation, equitable, premiums, payment, pensions, means-testing, lenders, savers, pensioners
Points of order	question, order, mr, put, speaker, deputy, point, grateful, read, agreed, record, time, minutes, may, call, standing, correct, apologise, madam, interventions	speaker, mr, madam, question, forthwith, deputy, apologise, order, o’clock, read, minutes, adjourned, accordingly, interventions, hansard, tomorrow, grateful, misled, correct, courtesy
Issues	important, issue, can, issues, take, ensure, hope, need, matter, consider, possible, place, also, concerns, deal, particular, course, taken, concern, raised	issues, issue, important, concerns, consider, possible, discuss, concern, particular, matter, considering, carefully, assure, understand, extremely, raised, addressed, obviously, address, expressed
Constituencies	many, constituency, constituents, problems, welcome, particularly, people, often, hard, face, others, feel, especially, country, worked, pay, concerned, represent, thousands, large	many, constituents, problems, hard, mine, difficulties, worked, represent, faced, feel, constituencies, thousands, hundreds, face, greatly, often, constituency, especially, worried, particularly
Ethnic groups & racism	action, taking, community, steps, taken, communities, take, actions, society, prevent, faith, groups, minority, church, black, ethnic, religious, freedom, race, diversity	religion, faiths, sikh, steps, racial, faith, sikhs, religious, priests, synod, beliefs, church, racism, taking, action, anglican, ethnic, hate, clergy, hatred
Amendments	clause, amendment, amendments, new, lords, section, 1, tabled, 2, clauses, line, 3, leave, act, shall, move, beg, 4, page, schedule	insert, nos, subsection1, amendmenta, amendment, subsection5, 1a, schedule, amendmentsa, amendments, subsection2, subsection6, clause, tabled, paragrapha, subsection, subsection3, andc, paragraphb, clauses
Reports	year, since, report, number, figures, official, march, april, published, 1997, figure, statistics, 15, 30, show, january, 2010, july, june, december	vol, october, march, official, february, july, january, november, june, april, 2011, statistics, since, 2009, 2007, december, 2005, figures, 2013, figure
People	people, want, get, one, go, can, think, see, need, know, say, things, much, like, good, problem, going, done, something, put	things, get, something, go, lot, want, talking, thing, trying, talk, think, really, quite, bit, else, happen, away, getting, enough, idea
Wales & Scotland	wales, scotland, scottish, england, welsh, assembly, parliament, devolution, uk, devolved, government, powers, kingdom, national, english, united, glasgow, executive, snp, edinburgh	scotland, scottish, welsh, snp, scotland’s, cymru, barnett, plaid, perth, wishart, holyrood, perthshirepete, wales, snp’s, assembly, devolved, dundee, scots, devolution, calman
Alcohol & tobacco	food, industry, alcohol, licensing, products, smoking, shops, shop, tobacco, advertising, health, standards, pub, pubs, high, buy, drinking, supermarkets, problem, retailers	tobacco, pubs, gambling, betting, labelling, drinks, cigarettes, casinos, smokers, cigarette, groceries, lap-dancing, vending, drinkers, supermarkets, fluoride, smoking, pubcos, retailers, pub
Place names	thank, south, north, constituency, excellent, join, congratulate, manchester, area, yorkshire, north-west, reply, visit, greater, visited, also, bristol, giving, nottingham, region	thank, wrexham, reddish, tameside, congratulating, newport, yorkshire, stockport, blaenau, derbyshire, south, north-west, stoke-on-trent, denbighshire, denton, nottingham, bristol, newingtonms, welcoming, congratulations
Budget	million, budget, year, billion, cuts, chancellor, spending, cut, increase, money, government, 1, funding, extra, next, investment, deficit, financial, crisis, growth	deficit, obr, billion, spending, budget, real-terms, forecast, million, borrowing, cuts, gdp, chancellor, cut, 2.5, chancellor’s, forecasts, 2010-11, 1.2, 1.5, finances
Tax	tax, pay, rate, income, wage, families, minimum, living, low, poverty, working, vat, increase, government, paid, national, paying, credits, average, poorest	tax, millionaires, 50p, vat, taxes, credits, wage, taxation, avoidance, incomes, rate, zero-hours, wages, 45p, earning, revaluation, income, richest, earners, regressive
Private companies	companies, company, market, financial, industry, competition, consumers, interest, consumer, assets, services, profits, markets, ownership, regulator, share, corporate, interests, customers, societies	mutuals, shareholders, provident, company, companies, competition, profits, corporate, shares, societies, company’s, co-operative, fsa, profit, co-operatives, directors, rock, regulator, assets, asset
Environment & fishing	environment, sea, fishing, marine, fisheries, industry, natural, fish, port, environmental, water, ports, rural, coastal, protection, conservation, fishermen, areas, management, area	fishing, fisheries, fishermen, cod, seas, whitby, coastguard, broads, cfp, angling, seafarers, anglers, inshore, discards, mmo, under-10, sssis, dredging, cockle, aonbs
Crime	crime, behaviour, victims, offence, criminal, serious, abuse, offences, antisocial, home, use, measures, drugs, drug, enforcement, offenders, problem, tackle, law, justice	sentences, asbos, cannabis, antisocial, offences, offence, trafficking, gangs, behaviour, penalty, sentencing, sentence, theft, criminals, custodial, offending, knife, heroin, offenders, convicted
Bills	bill, legislation, act, new, powers, provisions, regulations, power, place, provision, duty, apply, statutory, necessary, allow, provide, set, already, introduce, require	provisions, bill, bill’s, definition, legislation, regulations, statutory, passage, seeks, requirement, drafted, draft, statute, intention, safeguards, purpose, consult, legislative, amend, covered
Children	children, child, parents, families, children’s, support, poverty, family, young, needs, parent, start, adoption, adults, vulnerable, early, contact, must, need, autism	autism, csa, looked-after, adoptive, child, adopters, children’s, autistic, cafcass, nspcc, children, child’s, parent, dyslexia, adoption, kinship, childcare, intercountry, parents, lone
Utilities & PFI	public, private, sector, money, costs, cost, risk, value, management, service, water, government, contracts, contract, system, audit, flood, systems, agency, taxpayer	id, flood, nao, ofwat, public, contracts, auditor, private, purse, contractors, audit, pac, pfi, flooding, floods, taxpayer, contract, comptroller, defences, tendering
Middle East	security, government, peace, war, foreign, people, iraq, terrorism, international, conflict, threat, support, must, un, military, syria, israel, resolution, terrorist, refugees	syria, israel, palestinian, israeli, gaza, palestinians, syrian, saddam, arab, hamas, saudi, daesh, palestine, isil, israelis, hussein, lebanon, atrocities, assad, two-state
Local authorities	local, authorities, council, authority, areas, government, funding, area, councils, communities, county, grant, planning, community, central, formula, borough, locally, level, resources	local, authorities, councillors, councils, authority, unitary, county, formula, grant, lga, localism, locally, swindon, allocations, allocation, deprived, council, parish, authority’s, deprivation
Elections	vote, political, parliament, electoral, election, elections, elected, parties, people, voting, referendum, democracy, register, system, registration, democratic, commission, party, votes, majority	electoral, voters, turnout, voter, all-postal, votes, vote, voting, polling, first-past-the-post, av, referendums, elections, unelected, registration, ballot, candidates, electors, electorate, elected
Debate	members, debate, speech, heard, today, hope, opportunity, speak, hear, chamber, great, wish, support, time, pleased, debates, sides, like, follow, subject	debate, speech, members, debates, speeches, speak, heard, listened, sides, debating, hear, speaking, tonight, pleasure, chamber, thoughtful, listening, afternoon, queen’s, cross-party
Transport	london, transport, rail, bus, services, line, network, travel, airport, train, air, service, passengers, trains, railway, station, east, capacity, passenger, heathrow	rail, bus, passengers, trains, passenger, heathrow, railways, fares, freight, crossrail, hs2, high-speed, runway, electrification, airlines, gatwick, caa, baa, sra, thameslink
Questions	whether, information, may, answer, asked, ask, questions, response, available, advice, received, data, know, press, written, letter, department, meeting, details, officials	answer, information, questions, answers, data, written, details, letter, write, ask, officials, answered, asked, whether, informed, press, website, correspondence, response, requests
Families	family, life, families, lives, constituent, death, home, people, told, case, one, man, died, lost, mrs, person, mother, day, marriage, suffered	husband, mum, daughter, married, constituent, mrs, son, mother, marriage, died, father, wife, same-sex, death, loved, dad, suicide, funeral, bereaved, sister
Health research	research, treatment, cancer, medical, disease, health, drugs, condition, can, use, drug, patients, screening, risk, also, conditions, evidence, group, diseases, diagnosis	screening, asbestos, tissue, embryos, cancers, hepatitis, genetic, prostate, epilepsy, cloning, pleural, fertilisation, embryo, embryonic, ivf, anaemia, embryology, piercing, hfea, bowel
Dispatch box	back, come, look, forward, bring, moment, coming, comes, side, later, brought, along, bringing, round, looking, box, see, putting, sit, dispatch	come, back, look, moment, coming, forward, dispatch, comes, side, box, oh, surprise, bring, round, hoping, bringing, sooner, straight, along, sit
Parties	government, labour, conservative, party, opposition, policy, previous, liberal, conservatives, government’s, support, election, tory, front, democrats, coalition, benches, policies, general, fact	conservative, conservatives, liberal, democrats, lib, tory, democrat, benches, tories, opposition, manifesto, party’s, labour, benchers, dem, opposition’s, front-bench, party, spokesman, bench
Statements	us, said, just, let, say, now, tell, says, yet, saying, told, know, going, nothing, wrong, even, wants, words, minister’s, today	tell, says, let, wants, actually, us, saying, minister’s, telling, truth, wrong, wonder, thinks, nothing, promise, afraid, mistake, blame, admit, honest
European Union	european, eu, europe, union, countries, uk, britain, trade, single, british, negotiations, market, economic, france, germany, country, leave, membership, referendum, world	euro, ttip, brexit, accession, eu, currencies, cypriots, european, eurozone, europe, enlargement, pro-european, spain, currency, esm, france, greece, italy, brussels, isds
Locations	member, west, east, north, birmingham, friends, st, spoke, hull, sheffield, talked, leeds, leicester, midlands, upon, newcastle, westmr, eastmr, northmr, southmr	kingston, eastmr, bromley, chislehurstmr, holborn, dorsetmr, northmr, enfield, hull, southmr, chislehurst, stuart, ealing, rees-mogg, leicester, chingford, westmr, southend, greenmr, letwin
Jobs & manufacturing	jobs, economy, economic, growth, industry, unemployment, investment, government, uk, manufacturing, future, sector, employment, country, job, long-term, steel, north-east, industries, recession	steel, manufacturing, jobs, tata, teesside, economy, unemployment, recession, automotive, downturn, steelworkers, productivity, growth, inward, industries, recessions, nissan, economic, steelworks, double-dip
Small business	business, small, businesses, regulation, rates, enterprise, government, finance, support, firms, help, innovation, measures, smaller, regulatory, large, lending, enterprises, burden, larger	smes, medium-sized, businesses, business, enterprises, small, regulation, commerce, enterprise, entrepreneurs, tape, firms, lending, burdens, brs, start-ups, start-up, entrepreneurial, smaller, lend
Agreement & disagreement	believe, however, one, might, accept, must, different, case, system, view, change, think, whether, position, argument, rather, simply, reason, basis, although	accept, argument, view, principle, arguments, reason, might, argue, suggest, perfectly, balance, believe, suggesting, different, reasons, necessarily, sensible, disagree, argued, whatever
Voluntary sector	work, people, young, support, help, can, working, organisations, role, voluntary, ensure, together, good, also, need, important, encourage, opportunities, experience, community	voluntary, charities, organisations, volunteering, young, charity, youth, work, opportunities, helping, encourage, volunteers, encouraging, play, charitable, working, help, ways, valuable, together
Comments	member, said, shall, mentioned, earlier, points, lady, comments, referred, learned, intervention, remarks, interesting, raised, pointed, perhaps, gave, say, refer, described	comments, remarks, lady, interesting, points, happily, southwark, referred, bermondsey, referring, somerton, intervention, shall, intervened, mentioned, pointed, learned, earlier, gentlemen, rushcliffemr
Social care	care, services, social, carers, people, need, service, needs, support, older, provision, provide, quality, home, centres, elderly, access, provided, providers, homes	carers, hospices, dentists, dental, care, dementia, hospice, dentistry, respite, carer, advocacy, older, elderly, caring, palliative, dentist, milton, social, keynes, cared
Time	years, time, last, two, one, first, now, three, past, week, months, next, ago, every, 10, five, four, weeks, days, six	years, three, two, last, months, ago, past, time, four, week, weeks, six, five, first, next, days, 10, seven, half, now
Media & animals	bbc, farmers, digital, television, internet, animals, animal, media, radio, dogs, licence, dog, news, ban, farming, welfare, hunting, fee, online, farm	bbc, dogs, hunting, cull, bbc’s, badgers, badger, bovine, switchover, broadcasters, gm, fur, mink, poultry, circuses, analogue, hare, hounds, puppies, swine
Other	given, can, aware, may, recently, across, welcome, fact, government, well, take, close, result, seeking, indeed, support, responsible, confident, know, including	given, aware, can, recently, may, across, close, fact, welcome, seeking, confident, result, well, take, responsible, indeed, keep, regret, far, reconsider

3.6.2 Manual Validation

As STM is an unsupervised model, we used several different validation strategies to ensure the topics themselves are both interesting and relevant (Grimmer & Stewart, 2013). Quinn, Monroe, Colaresi, Crespin, & Radev (2010) suggest that topics are valid if they correspond to external events. Figure shows the number of speeches by Labour MPs on the “Middle East” topic, with a spike in 2003 (at the start of the Iraq War), another spike in 2008 and 2009, as the bulk of British troops left Iraq, a small spike in 2011 coinciding with UK participation in NATO’s military intervention in Libya, and debate in 2014–2016 over UK participation in military interventions in the Syrian Civil War.

Figure shows debate over the devolved authorities of Wales and Scotland peaking in 2014, to coincide with Scotland’s independence referendum. The post-2015 decline also likely stems from the SNP winning all but three seats in Scotland during the 2015 General Election. Figure shows the increase in debate over the European Union coinciding with the referendum on the UK’s member of the European Union.

Number of Speeches in “Middle East” Topic per Year

Number of Speeches in “Wales & Scotland” Topic per Year

Number of Speeches in “European Union” Topic per Year

3.6.3 Topic Proportion

Figure displays the percentage of speeches made by AWS Labour MPs compared to non-AWS female Labour MPs, for all topics where AWS and/or non-AWS MPs made 100 or more speeches.

k0 Topic Proportions

“Benefits & disability” is the topic that AWS MPs are proportionally most likely to discuss, relative to their non-AWS colleagues; AWS MPs are 2.09 more likely to discuss “Benefits & disability” than non-AWS Labour MPs. Non-AWS female Labour MPs are most disproportionately likely to discuss the “Legal system” topic, with -1.45 as many of their speeches covering the “Legal system” than AWS MPs.

5.1 AWS References to Constituents in Context

A random selection of 2% of all references to “my constituency”, “my constituent” and “my constituents”, by AWS MPs, in context.